Zhihao Chen comments

Results 33 comments of


                                            Zhihao Chen

How to remove the shadow from the original image?

In our work, we just focus on shadow detection rather than shadow removal.

Some error in your shadow detection results at test datasets

Q1: We generate the outputs again and get the similar results. The reason of the comparability of SBU and SBU_crf maybe lead by that we use the binarization operation (prediction...

Some error in your shadow detection results at test datasets

Actually, we just do this binary operation in SBU. For UCF and ISTD, we save the soft output before crf.

error

In 'train.py' line25, you can import the joint_transforms like "from dataloaders import joint_transforms_edge as joint_transforms"

test result

that we use the binarization operation (prediction = (prediction>90)*255) before CRF in SBU dataset. Because at that time, we observed that many pixels are positive but still under 127.5, we...

请问什么时候可以开放代码呢，万分感谢！

最近有点忙，大概8月份能整理出代码

关于“半监督”MT给网络带来的增益，有点疑惑，请教一下师兄

关于半监督为什么能提升模型训练精度这个问题。一方面，引入了更多的数据，增强了网络的泛化能力，一定程度上避免了数据在小数据集上的过拟合。另一方面，这种软标签的训练方式可以理解为一种正则化方式。这点也可以解释为什么很多无监督学习方法也能学习到数据分布特征了。

Some question about your code

yes, I'll sort it out when I'm free

Some question about your code

I release some unsorted code in 2020.9.17.

Where I can download your paper? The line you provice is invalide.

You can download it now